| Các trang trong chủ đề: < [1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16] > | Complimentary TM-Town membership for ProZ.com members Người gửi thông tin lên tuyến đoạn: Jared Tabor
|
|---|
..... (X)
Local time: 10:40
Tomas Mosler, DipTrans IoLET MCIL MITI
TM Town FAQ April 25: "TM-Town's directory rankings are not based purely on any one metric. Translators who have loaded aligned work (i.e. a TM or glossary) in a specific language pair will rank over translators who have only loaded monolingual documents or have not loaded work in that specific language pair."
(No reference to Deshi, and Deshi demo video seems to use a monolingual termbase.)
I wrote: "TM Town privileges (by better ranking) such use (sharing of aligned TUs) over less risky option (Deshi)."
Kevin wrote: "Loading a TM directly to TM-Town or loading a Deshi file of the TM has the exact same effect."
You wrote: "All this means is translators who upload bilingual data will rank higher than those who upload only monolingual data. Of course they will. That's the whole point of the system."
I wrote: "I said the very same thing about ranking."
(At that point I still understood - or misunderstood - that Deshi is monolingual.)
TM Town FAQ April 26: "TM-Town's directory rankings are not based purely on any one metric. Translators who have loaded aligned work (i.e. a TM, glossary, or Deshi-generated .town file) in a specific language pair will rank over translators who have only loaded monolingual documents or have not loaded work in that specific language pair."
That indeed puts Deshi on par with other inputs. But I'm still curious to know how Deshi produces aligned bilingual input; I'm not saying that it doesn't or cannot do it, it is only not clear from the demo video where the .town file at 1:09, opened in a text editor, contains only English words - where is Spanish? And if both languages are uploaded, shouldn't "Jörg" be redacted also in the other language? Or is this demo for a monolingual version of Deshi (if Deshi is not strictly bilingual)?
A Deshi .town file includes only terms from the source segments of the TM. When talking about one's ranking in the TM-Town directory or getting matched in Nakōdo, uploading a Deshi .town file is equivalent to uploading a TM.
Uploading a TM, glossary or .town file helps one's ranking more than uploading a monolingual document (i.e. only a source document or only a target document).
| | | | Willem Wunderink
Hà Lan
Local time: 02:40
Thành viên kể từ 2011
English to Dutch
+ ...
| Added value? I don't see it | Apr 27, 2016 |
To be honest I don't see any added value in this TM-Town system.
If I want work I use Proz.com (and competitors if necessary), if I want to search TM's I have my own database: why taking all the effort in uploading files etc.?
For me it's not worth the trouble and time.
| | | |
Willem Wunderink wrote:
... I want to search TM's I have my own database: why taking all the effort in uploading files etc.?
For me it's not worth the trouble and time.
Finally somebody who understands it. I was getting desperate. Holy faeces!
H.
| | | |
Hello,
I have just downloaded TM-Town Desktop Application 2.0.0 but an error message appears telling me to contact my provider  .
What can be done?
Thanks
| | |
|
|
|
..... (X)
Local time: 10:40
Hi Susana,
Are you using Windows? If so, please see the Windows Install Guide at the bottom of this page. On Windows, you will need to first install an application called Shoes which will help Deshi run. You can do... See more Hi Susana,
Are you using Windows? If so, please see the Windows Install Guide at the bottom of this page. On Windows, you will need to first install an application called Shoes which will help Deshi run. You can download it here.
Here is a video showing the full installation process on Windows.
If you still have trouble after trying the above, please let me know.
Kevin ▲ Collapse
| | | | | the weakest link | Apr 27, 2016 |
Kevin Dias wrote:
Tomas Mosler, DipTrans IoLET MCIL MITI wrote:
OK, but how does .town file get "aligned"? In the Deshi sample video on your website, there is only a monolingual collection of words. How it gets linked to the source or target language? How are the adequate/correct counterparts (words) in the other language selected? Or if "aligned" here means something else, then what is it?
The Deshi .town file is created from an aligned TM file. That is the meaning here.
OK, but FAQ says: "Translators who have loaded aligned work (i.e. a TM, glossary, or Deshi-generated .town file)..."
If Deshi is created from aligned TM, but is monolingual itself, then by definition it is not "aligned work", but rather something like "data derived from aligned work"?
Tomas Mosler, DipTrans IoLET MCIL MITI wrote:
unlike with TM Town, I'm not sending the whole material (bilingual, monolingual or TM) anywhere
This is the point you seem to be missing. You do not have to send the "whole material" anywhere with TM-Town. You can send a synopsis file created by Deshi which does not include any full segments or any aligned segment data.
I'm not missing that. However any solution is only as good as its weakest link.
I'm not concerned about what the user doesn't have to do. I'm concerned about what he can do. What choices TM Town gives him. As long as one choice may lead to confidentiality violation, and at the same time TM Town is promoted by ProZ which endorses specific guidelines, I see a problem.
First of all, not all TMs fall under a NDA or confidentiality agreement.
Of course. But (such exceptions aside) not having a verbatim NDA or not being bound by association rules doesn't automatically mean the client expects the translator to distribute source & target anywhere, does it? Further, to my understanding the translator owns only the target at best, but never the source.
Either way, I think it could be quite revealing to know how many translators who uploaded their own TMs (not something from the public domain like DGT) to TM Town, when asked "Did your client approve this?" or "If you tell your client now, are you sure they would not mind?", would answer "YES!" enthusiastically. (That is not to imply they would intentionally do something incorrect, rather that they maybe wouldn't think twice - after all, as somebody noted, they may trust ProZ which can in theory reduce the caution.)
For example, a translator could upload a public DGT TM file
This leads me to another concern. Wouldn't that be unfair in regards to ranking? I mean, surely everyone can take the DGT TM, but isn't the principle of TM Town to show my own skills? If someone personally translates 20,000 words for EU in public domain and uploads his TM to TM Town, he will have worse ranking (number of TUs clearly affects this) than someone who simply without any hesitation takes the whole DGT TM created by others... (Sure, there are terminology filters but not everyone uses them, and they would not help much in this scenario.)
What is the point of such ranking system then, when a better place can be achieved simply by collecting as much bilingual data on the Internet as possible?
| | | | ..... (X)
Local time: 10:40
| For Reference Only | Apr 27, 2016 |
Tomas Mosler, DipTrans IoLET MCIL MITI wrote:
For example, a translator could upload a public DGT TM file
This leads me to another concern. Wouldn't that be unfair in regards to ranking? I mean, surely everyone can take the DGT TM, but isn't the principle of TM Town to show my own skills? If someone personally translates 20,000 words for EU in public domain and uploads his TM to TM Town, he will have worse ranking (number of TUs clearly affects this) than someone who simply without any hesitation takes the whole DGT TM created by others... (Sure, there are terminology filters but not everyone uses them, and they would not help much in this scenario.)
What is the point of such ranking system then, when a better place can be achieved simply by collecting as much bilingual data on the Internet as possible?
You conveniently only quoted the first half of my sentence. Let me re-quote the full sentence:
"For example, a translator could upload a public DGT TM file, mark it as 'For Reference Only', and then search those segments through our browser search feature or through one of our CAT tool extensions."
As defined here:
This document is for reference only
Choose this option if you did not translate the document, but you would like to keep it on TM-Town for storage or reference purposes. Documents marked as "Reference Only" will not count toward your translation unit or term concept count.
| | | | ..... (X)
Local time: 10:40
| On Translation Quality / TM Verification | Apr 27, 2016 |
There have been some very good questions asked in this forum with respect to measuring the quality of the material that is uploaded to TM-Town. I've been meaning to answer those, so here goes.
Sheila Wilson wrote:
Is volume going to be master now? And how on earth can clients know whether there's any actual skill behind that volume?
Ivonne Reichard-Novak wrote:
Quality or quantity?
Bernhard Sulzer wrote:
who created it and non-validation (how good or bad is it really)) of content are two red flags
Jo Macdonald wrote:
About validation
I have read the post you directed Sheila to earlier in this thread but from my brief experience on your site it does actually look like it would be really easy for anyone to post any TM or other data they have on their system or got sent by a client and declare "I translated this" as really the only validating step I came across was the drop-down menu asking me if I had translated the work, or part of it, etc.
Mirko Mainardi wrote:
I found myself nodding... while reading Ivonne's "Quality or quantity?" post, as I do share her concerns
Erik Freitag wrote:
Other concerns are still valid, though, mainly the issue that the amount of translated work within a given context documented via a Deshi file doesn't say anything about the quality of the translations. The fraudulent workaround described in an earlier post (get large chunks of commercially interesting source texts -> machine translation -> upload TM or Deshi file to TM-town -> high ranking) is, although an extreme case, quite possible. But even if you replace the MT stage by substandard human translation, no client is going to know.
So, while I still think that there are quite some important problems to solve (and I'm not sure whether they can be solved at all), I'm willing to give it a try, since I won't risk anything using Deshi files.
Let me first explain how we are currently tackling the issue on TM-Town. Then I'd like to lay out some ideas we have to address the problem even better in the future (along with the pros and cons of those ideas). I would love to get everyone's feedback and ideas on the topic and hopefully we can have a meaningful discussion.
How things currently work on TM-Town
When a translator uploads a file to TM-Town, the translator is first asked to declare the document using one of the following options:
- I translated this document myself
- I partially translated this document
- I only proofread this document
- This document is for reference only
These options are explained as follows:
I translated this document myself
Choose this option if:
- The document is a source document which you translated
- The document is a target document and you did the translation
- The document is a translation memory file and you translated all of the target segments
- The document is a glossary that you created
I partially translated this document
Choose this option if you did not translate all of the document, but you did translate parts of it. Documents marked "Partially Translated" will still get limited credit in matching; however, these documents will not count toward your translation unit or term concept count.
I only proofread this document
Choose this option if you did not translate the document, but you did proofread it. Documents marked "Proofread" will still get limited credit in matching; however, these documents will not count toward your translation unit or term concept count.
This document is for reference only
Choose this option if you did not translate the document, but you would like to keep it on TM-Town for storage or reference purposes. Documents marked as "Reference Only" will not count toward your translation unit or term concept count.
Locked documents
TM-Town's system also analyzes documents for certain heuristics to detect public TMs and glossaries and may automatically classify a document as "Reference Only" and lock the classification. Additionally, TM-Town will manually lock a classification as "Reference Only" if we have evidence that the document was improperly classified.
Over time this system will improve as our Nakōdo technology allows us to create a unique "fingerprint" of a document, which we can then analyze for similarity against finger prints of known public documents or documents that have been uploaded by other translators on TM-Town.
The above steps are not a catch-all by any means. It is very possible to "trick" the system and claim a document as your own, when it is not.
In addition to what is described above, we also look for certain red flags. We look for outliers in translation unit volume numbers relative to other profile fields such as years of experience, etc. For example, a translator with a high number of uploaded translation units but 1 year of experience would raise a red flag. Translators who have a high number of uploaded translation units but do not have any ProZ WWA positive reviews or do not have any KudoZ points or are not Certified PRO members may also raise potential red flags.
Still, given the above, it is possible for an "impostor" to slip through the cracks. Which brings me to what is most likely our most powerful means of detection.
TM-Town is self-policing
No offense, but translators would make for terrible mobsters. Translators love to rat each other out 
All joking aside though, this is actually a very effective means of controlling this issue on TM-Town. Professional translators with many years of experience in their field(s) and language pairs tend to have a good feel for who their competition is. They have seen the names of others in their niche. Translators are very often searching their areas in the TM-Town directory to see where they stack up. If they suddenly see a name and face they don't recognize that has a huge number of translation units it rings alarm bells for them and they will be very quick to bring it to our attention and where we can then look into it further and take any appropriate action if necessary.
To recap, as things currently stand on TM-Town there are systems and policies in place to identity issues of misuse and take appropriate action. While an inexperienced translator may be able to "fake" it to the top, they won't stay there for very long.
However, the above doesn't address the issue of quality. A translator may have 10 years of experience, which in reality may just be 1 year of experience 10 times. The documents s/he uploads are 100% that person's work...but the quality of said documents might be terrible. As such, a translator might look good on TM-Town as having uploaded a lot of work, but in reality be a poor translator.
So how are we currently tackling this quality issue?
Currently we take no stand on the quality of any uploaded work. Instead what we aim to do is give a client or visitor browsing a translator's profile on TM-Town other criteria and heuristics from which they can gather evidence of that translator's ability to produce a quality translation. These include:
- Years of experience
- Education
- Associations and Memberships
- Certified Pro Status
- KudoZ points
- WWA feedback
- CAT tools used
- Links to other translation portal profiles
- Sample translations
Nakōdo can tell the client the translators who have translated similar material in the past, but ultimately it is up to the client to evaluate that translator's profile to decide if s/he has the appropriate level of experience and expertise to do the job at the quality level that client is searching for.
Where do we go from here?
In an ideal world, when a document is uploaded we would be able to give that document a quality score that accurately reflects the quality of translation.
OK, sounds simple on the surface, but let's explore some of the issues associated with such a system and discuss the different pros and cons.
Issue #1 - Privacy
There will never be a fully automatic way to judge the quality of a translation. Therefore, it is necessary that someone, a human, look at the translation in question to judge the quality. This would be in conflict with our current privacy policy as all documents uploaded to TM-Town are completely private. We would need to carve out a way that a TM-Town member could give permission to another peer to grade their work. It is most definitely possible to solve this problem and still maintain privacy and confidentially; however, when designing or thinking about a potential solution, this needs to be the first item to consider.
Issue #2 - Past Performance is Not Necessarily Indicative of Future Results
A translation that a translator completed 10 years ago, may be at an entirely different quality level than that translator can produce now. That translator now has 10 additional years of experience and knowledge, that if they translated that same material again today the quality level would be entirely different. Do we want to penalize that translator and create a disincentive for them to upload that document? Nakōdo could match them to a client who needs a translation in the same area of that document the translator completed 10 years ago, but today that translator can complete it at a much higher quality level.
How should this be handled? Should there be a type of anti-decay function that the quality score of a document rises as it ages and the translator gains more experience and expertise? Should one's overall quality score be more highly weighted to the most recently translated material?
Issue #3 - Translators (generally speaking) do not like being scored
It is human nature, but any score less than perfect on your translation and you are not going to be happy. Just do a search for proposals of negative WWA on the ProZ forums and you will see many concerns. Are you prepared to have your translations scored for quality?
Issue #4 - Translation is an art, not a science
One translator's idea of a great translation may be another translators 'eh, it's just OK' translation. There is rarely only one "right" answer. On the other hand, there are wrong translations, there are grammar and punctuation mistakes, there are misuses of terms. I don't think this issue is a show stopper, but is something to keep in mind.
OK, so with those issues in mind what are some potential solutions?
One idea would be to use the same concept that ProZ uses for translation contests. Did you know that ProZ has a patent on this? It is a very interesting system and works very well to determine the quality (or relative quality) of a translation. We could do something similar. The good thing about this type of system is that we could anonymize the translator of the text. Reviewers would never have to review a full TM (impossible and unecessary) but instead just a random statistical sampling of segments from the TM.
Those are my thoughts on the quality issue. I'd love to hear feedback from the community on this and if you have any other ideas or thoughts.
| | |
|
|
|
Kevin Dias wrote:
Hi Susana,
Are you using Windows? If so, please see the Windows Install Guide at the bottom of this page. On Windows, you will need to first install an application called Shoes which will help Deshi run. You can download it here.
Here is a video showing the full installation process on Windows.
If you still have trouble after trying the above, please let me know.
Kevin
Yes, I'm using Windows 10. I will do this, thanks a lot.
| | | | Tom in London
Vương Quốc Anh
Local time: 01:40
Thành viên kể từ 2008
Italian to English
Kevin Dias wrote:
a lot
It all sounds like a translators' dating agency. With the same accuracy.
| | | | | Pushing files | Apr 28, 2016 |
One question, is it possible to "push" sample translation files to TM-Town or the translator will need to recreate the files they have on their ProZ profile?
The same question applies to glossaries created in ProZ profile.
| | | | ..... (X)
Local time: 10:40
Merab Dekano wrote:
One question, is it possible to "push" sample translation files to TM-Town or the translator will need to recreate the files they have on their ProZ profile?
The same question applies to glossaries created in ProZ profile.
Yes, it is possible to push your ProZ.com sample translations to TM-Town, please see this demo video.
For glossaries, it is not possible at this moment but we are working on making that possible in the near future.
| | |
|
|
|
| response to response | Apr 28, 2016 |
Kevin Dias wrote:
Maxi Schwarz wrote:
My expertise (in a subject area) has already been determined by my professional body. If I want my area of expertise listed, I need to offer proof in the sense that I might have a degree and/or experience in that area. My examples were a doctor or nurse for expertise in medicine, my own B. Ed. and teaching experience for expertise in education. The professional body looks at and verifies this background as knowledgeable thinking human beings, and determine that the translator does indeed have that expertise.
Nothing listed above does the same.
Listed above were both Education and Associations and Memberships. In other words, on the TM-Town profile there are 2 very clear areas where a translator can list one's degrees and professional bodies s/he belong to.
This does not pertain to what I was writing about. "Expertise" being "determined" by uploaded translations in a subject area (TMT) on the one hand. Expertise having actually been DETERMINED by a professional body which has examined background.
| | | |
Having looked at things posted since the last time.
My first concern is about the idea "ranking" translators who join. I find that idea quite disturbing. That would give clients the impression the idea that some translators are "better" than others - or more damaging, "worse" than others. A client may choose one over the other based on such ranking, believing that it is validity. There is also the matter of reputation, which we spend years building. The ranking itself is done v... See more Having looked at things posted since the last time.
My first concern is about the idea "ranking" translators who join. I find that idea quite disturbing. That would give clients the impression the idea that some translators are "better" than others - or more damaging, "worse" than others. A client may choose one over the other based on such ranking, believing that it is validity. There is also the matter of reputation, which we spend years building. The ranking itself is done via these uploaded sample translations, with the problematics of that already discussed here and there.
Further on I read a longer discussion on the idea of quality under "issue number 4". The long and short of what is written under issue 4 is the opinion that quality cannot be determined. And yet translators are to be ranked! The number one important factor of a translation is quality and many of us base our reputations on things like quality and reliability (go hand in hand). I would not want to rank translators in the first place, but if I don't think that quality can be determined, I definitely would not try to do so. After discussing how quality can't be determined there is an idea of maybe using something like contests with anonymous participants reviewing documents - this within the guise of a profession? It still goes back to the same thing - the idea of ranking translators itself.
Issue #3 - Translators (generally speaking) do not like being scored
It is human nature, but any score less than perfect on your translation and you are not going to be happy. Just do a search for proposals of negative WWA on the ProZ forums and you will see many concerns. Are you prepared to have your translations scored for quality?
The "human nature" bit makes it all seem childish and unprofessional. My work has been examined - by a professional body based on well thought out criteria, and examined by knowledgeable experienced professionals. I don't mind that kind of thing at all. In fact, I have acted as examiner for certification myself for a number of years. An important point: we do not give a score or a rank / simply a pass or fail - and for good reason. It would be the height of hubris to think that you can determine one translator as being "better" than another, and so it isn't done. If I wouldn't do it as an examiner, if it was not done by my professional body, then I do not want it to be done by some site that invites me to add my name to their ranks. Because the idea is quite arbitrary.
For the second part, I am not that familiar with the WWA's but have seen a similar thing elsewhere. People will rate others for the wrong reasons, or lack of knowledge. If someone is going to assess my work, then I want that person to be knowledgeable in my field.
Think of the on-line student ranking of their teachers. You'll see a teacher get a high rank because he is "nice" and passes all students with a high grade, while a sincere teacher may get a low feedback because he is trying to teach to high standards. Not wanting to be subject to such a thing has to do with common sense and not human nature.
For the majority of translation jobs that translators do on a daily basis, 100% of the translation IS NOT confidential (of course there could be exceptions such as government jobs, medical records, etc.). ...
This statement surprised me, and I was trying to picture a non-confidential type of translation. That said, the things listed under "of course there could be exceptions" lists the majority of the translations that I do.
The statement on confidentiality continues:
A typical NDA states that you are not to disclose any Confidential Information. With Confidential Information being a legal defined term in the document, which as a translator signing that document you should have a good idea of what it means. If you don't, you should seek clarity from the client before signing the NDA.
Kevin, you may not be aware of it, but for many of us, confidentiality does not involve NDA's - it is in the code of ethics of our profession, period. I signed a code of ethics with my professional organization when I was awarded certified status, and confidentiality is in there. Even if it were not, my own personal ethics would not allow me to violate confidentiality regardless of whether a client stipulates it in an NDA. The majority of my clients do not have me sign an NDA. They simply assume that those kinds of standards exist.
This:No offense, but translators would make for terrible mobsters. Translators love to rat each other out
All joking aside though, this is actually a very effective means of controlling this issue on TM-Town. Professional translators with many years of experience in their field(s) and language pairs tend to have a good feel for who their competition is. They have seen the names of others in their niche. Translators are very often searching their areas in the TM-Town directory to see where they stack up.
Actually that is both wrong and offensive. I don't think I'm alone in saying that I spend my professional time responding to clients, doing translations, researching the material that gets translated. I do not spend time trying to figure out what other translators are doing, judging them, "ratting out" translators (to what authority? I only know of my own professional organization). The idea of seeing "where I stack up" by looking at a directory has never crossed my mind. If I want to know "where I stack up" then it is based on new clients approaching me because of recommendations by my customers, repeat orders by customers, and feedback given by customers.
That said, I will observe what colleagues write, how they write it, how they help in terminology questions and such, in view of collaboration with such colleagues. From time to time I have asked a colleague to review my work, when the subject area falls into their specialization. I will go not only by signs of their expertise in an area (how they answer a terms question, plus background), but also maturity, responsibility, and attitude. It is not anything as childish as wanting to "rat out" or compare myself to others. I imagine that I'm not alone. ▲ Collapse
| | | | Krys Williams 
Vương Quốc Anh
Local time: 01:40
Thành viên kể từ 2003
Polish to English
+ ...
| No thank you | Apr 28, 2016 |
Well I finally got in, but got stuck at the point where the web site wanted me to upload translations or a TM.
Sorry, my translations and TMs are my intellectual property and I am not making them available for others to steal.
So then it suggested I download some unknown software which would "analyse" my TMs.
I searched for this "Deshi" software online and found no pertinent references apart from a link to a dodgy Indian site. No way am I going to instal... See more Well I finally got in, but got stuck at the point where the web site wanted me to upload translations or a TM.
Sorry, my translations and TMs are my intellectual property and I am not making them available for others to steal.
So then it suggested I download some unknown software which would "analyse" my TMs.
I searched for this "Deshi" software online and found no pertinent references apart from a link to a dodgy Indian site. No way am I going to install something like that!
This all seems very much like a scam.
I have more than enough work offered to me and new client enquiries as it is, so I am not willing to engage with such a dodgy sounding web site just on the offchance of getting another client or two!
Actually, the more I think about this, the angrier I get that Proz is pushing it. I am seriously considering cancelling my Proz membership as a result.
[Edited at 2016-04-28 19:26 GMT] ▲ Collapse
| | | | | Các trang trong chủ đề: < [1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Complimentary TM-Town membership for ProZ.com members | Pastey | Your smart companion app
Pastey is an innovative desktop application that bridges the gap between human expertise and artificial intelligence. With intuitive keyboard shortcuts, Pastey transforms your source text into AI-powered draft translations.
Find out more » |
| | Trados Studio 2022 Freelance | The leading translation software used by over 270,000 translators.
Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop
and cloud solution, empowering you to work in the most efficient and cost-effective way.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
